Hypothesis

Aging brain actively removes metabolically inefficient, poorly connected neurons via complement‑tagged microglial phagocytosis, a process analogous to the clonal expansion of mutant stem cells in the aging colon but opposite in outcome: neuronal eviction preserves circuit efficiency at the cost of cell number, whereas colonic field cancerization permits deleterious clones to expand.

Mechanistic Model

Neuronal energy status regulates surface expression of the “don’t‑eat‑me” signal CD47 through SIRT1‑dependent deacetylation. Declining NAD+/SIRT1 activity in aged neurons reduces CD47, while heightened mitochondrial ROS increases extracellular ATP and exposes calreticulin, together promoting C1q and C3 deposition. Microglia expressing CR3 phagocytose these tagged cells, clearing synaptic weak links. In contrast, colonic stem cells lack efficient microglial surveillance; mutant clones evade immune detection due to tolerant niche signals (e.g., TGF‑β, IL‑10) and accumulate mutations, driving field cancerization 1 2. Thus, the brain employs a quality‑control pruning mechanism that the renewing gut epithelium does not.

Testable Predictions

1. Older mice will show elevated C1q/C3 binding specifically on neurons with low mitochondrial membrane potential (measured by TMRM) and reduced synapse density.
2. Genetic or pharmacological boosting of NAD+/SIRT1 in neurons will increase CD47 levels, decrease complement deposition, and rescue neuron numbers without impairing behavioral performance.
3. Blocking microglial CR3 will prevent neuron loss in aged brains but lead to accumulation of hypoconnected neurons and deficits in learning tasks.
4. In human post‑mortem cortex, cortical layers with high SFRP2 methylation (a field cancerization marker) will not show corresponding neuronal eviction signatures, highlighting tissue‑specific divergence.

Experimental Approach

• Use aged (24‑month) Cx3cr1‑GFP mice; isolate neurons by FACS based on TMRM low/high and perform flow cytometry for C1b, CD47, calreticulin.
• Treat cohorts with NR (nicotinamide riboside) to raise NAD+ or with SIRT1 activator SRT2104; assess neuron numbers (NeuN+ density) and complement deposition via immunohistochemistry.
• Apply anti‑CR3 antibody or CR3‑KO mice; quantify surviving neurons, synaptic marker (PSD‑95) puncta, and performance in Morris water maze.
• Parallel human dataset: compare DNA methylation arrays from colon and cortex (TCGA, GTEx) for SFRP2, EVL/miR‑342 and correlate with neuronal density markers (NEUN) from neuropathology scores. Statistical tests: two‑way ANOVA with age and treatment as factors; significance set at p<0.05.

If predictions hold, the data would support a conserved pruning logic driven by metabolic insufficiency, challenging the view that neuronal loss is merely passive damage and instead framing it as an adaptive, albeit potentially maladaptive, response to energetic decline.